---
title: The AI Act for companies that only use AI | Disruptive Lab
description: A company that uses AI but builds none is a deployer under the AI Act. What applies already, what comes in December 2027 and what to keep on record.
url: https://disruptivelab.dev/en/news/ai-act-for-deployers
language: en
---

1.  [News](https://disruptivelab.dev/en/news)
2.  [European rules](https://disruptivelab.dev/en/news/european-rules)
3.  The AI Act for companies that only use AI

[Guide](https://disruptivelab.dev/en/news/guides)  [European rules](https://disruptivelab.dev/en/news/european-rules)

# The AI Act for companies that only use AI

What the EU AI Act, as amended in July 2026, asks of a company that buys and uses AI, and from when.

- **Published:** 17 November 2026
- **Updated:** 17 November 2026
- **Reading time:** 8 min read
- **Editor:** David Lahuta

## In short

-   A company that uses AI under its own authority and builds none is a "deployer" under the EU AI Act.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)
-   Two duties have applied since February 2025. Staff who work with AI must be helped to understand it (AI literacy), and a short list of practices is banned.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4)
-   The transparency rules for deployers, such as labelling deepfakes, have applied since 2 August 2026.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4)
-   The duties for high-risk uses, such as AI that screens job applicants, apply from 2 December 2027 after the amendments of July 2026.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)

## Contents

1.  [Are we a provider or a deployer?](https://disruptivelab.dev/en/news/ai-act-for-deployers#are-we-a-provider-or-a-deployer)
2.  [What has applied since 2025?](https://disruptivelab.dev/en/news/ai-act-for-deployers#what-has-applied-since-2025)
3.  [What changed on 2 August 2026?](https://disruptivelab.dev/en/news/ai-act-for-deployers#what-changed-on-2-august-2026)
4.  [Which of our uses could be high-risk?](https://disruptivelab.dev/en/news/ai-act-for-deployers#which-of-our-uses-could-be-high-risk)
5.  [Who oversees the AI, and what is recorded?](https://disruptivelab.dev/en/news/ai-act-for-deployers#who-oversees-the-ai-and-what-is-recorded)
6.  [What will an auditor or supervisor ask to see?](https://disruptivelab.dev/en/news/ai-act-for-deployers#what-will-an-auditor-or-supervisor-ask-to-see)

## Are we a provider or a deployer?

The AI Act, Regulation (EU) 2024/1689, sorts companies by what they do with AI. A provider develops an AI system, or has one developed, and offers it under its own name or trademark. A deployer is any person, company or authority that uses an AI system under its authority, unless the use is purely personal.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

A firm that buys an AI assistant for its staff or switches on an AI feature in its accounting or HR software is a deployer. Most of the Act is written for providers. A deployer's list is much shorter.

The line can move. A deployer is treated as the provider of a high-risk system, with the provider's far heavier duties, if it puts its own name or trademark on such a system, makes a substantial change to one, or changes the intended purpose of an AI system so that it becomes high-risk.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) Taking a general AI assistant and setting it to rank job applicants is the kind of change of purpose worth noticing before it happens.

## What has applied since 2025?

Two parts of the Act have applied since 2 February 2025, AI literacy and the bans.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4)

**AI literacy (Article 4).** AI literacy means the skills, knowledge and understanding to use AI in an informed way and to be aware of its opportunities, risks and possible harm.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) The original Article 4 asked companies to ensure "a sufficient level of AI literacy" of their staff.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) The Digital Omnibus on AI, Regulation (EU) 2026/1744 of 8 July 2026, replaced that article. It was published in the Official Journal on 24 July 2026 and came into force three days later.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)

The new wording keeps the duty with the company. Providers and deployers "shall take measures to support the development of AI literacy" of their staff and of others who operate or use AI on their behalf, taking into account their knowledge, experience and the context of use. The duty does not require any specific level of AI literacy in any one person. The Commission and the member states must support companies in meeting it, SMEs in particular.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)

The Commission's own guidance is practical. No certificate is needed, there is no single format, and a company can keep an internal record of its trainings and other guidance.[3](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-3) Asking staff only to read the instructions for use may not be enough. A company whose people use a chat assistant to write advertising texts or translate should tell them about specific risks such as hallucination, which means answers the AI makes up.[3](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-3) According to the Commission, national market surveillance authorities supervise and enforce these rules from 2 August 2026.[3](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-3)

**The bans (Article 5).** A few uses of AI are forbidden outright. Among them are manipulating people with deceptive techniques that cause significant harm, exploiting people's age, disability or economic situation, inferring the emotions of people at the workplace or in education except for medical or safety reasons, and sorting people by biometric data to deduce their race, political opinions, religion or sexual orientation.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) An ordinary employer is most likely to meet the ban on reading emotions at work, for example in a tool that scores the mood of call centre staff from their voice. A breach can cost up to EUR 35 million or 7% of worldwide annual turnover, whichever is higher, and for an SME whichever is lower.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

The Omnibus added two bans that apply from 2 December 2026. They cover AI that generates realistic intimate images of an identifiable person without that person's explicit consent, and AI that generates child sexual abuse material. For a deployer, use is banned where it uses the system for that purpose.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4)

## What changed on 2 August 2026?

Most of the Act has applied since 2 August 2026, including the transparency rules of Article 50.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4) For a deployer they mean three things.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

-   People exposed to emotion recognition or biometric categorisation must be told the system is in use.
-   An image, sound or video made or altered by AI that resembles real people, places or events and would falsely appear authentic, a deepfake, must be disclosed as such.
-   Text made or altered by AI and published to inform the public on matters of public interest must be disclosed as AI made, unless a person reviewed or edited it and a person or company holds editorial responsibility for it.

A chat window on your website is first the provider's matter. The provider must design a system that talks to people so that they know they are dealing with AI, unless that is obvious.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) It is worth asking your supplier who the provider is and how this is done.

The Omnibus did not delay any of these deployer duties. It gave only the providers of generative AI already on the market before 2 August 2026 until 2 December 2026 to add machine-readable marking to what their systems produce.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2) A breach of Article 50 can cost up to EUR 15 million or 3% of worldwide annual turnover, again whichever is lower for an SME.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

## Which of our uses could be high-risk?

A high-risk use is one listed in Annex III of the Act, where AI can affect people's lives, health or rights. For an ordinary company the area to check first is employment. It covers AI used to recruit or select people, to place targeted job adverts, filter applications and evaluate candidates. It also covers AI used to decide on promotion, termination or terms of work, to allocate tasks on the basis of behaviour or personal traits, and to monitor and evaluate staff performance.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) Annex III also names AI that assesses the creditworthiness of individuals and AI that prices life and health insurance.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

A use on the list is not high-risk where it poses no significant risk, for example because it does not materially influence the outcome of a decision. Article 6(3) sets the conditions.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

The Omnibus moved the dates. The rules for Annex III uses now apply from 2 December 2027, and those for AI built into products covered by the EU product laws in Annex I from 2 August 2028.[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)[4](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-4)

A typical use

What the Act makes of it

Since or from

Staff draft, translate or summarise with a chat assistant

Not high-risk in itself. AI literacy applies

2 February 2025

A tool that reads employees' emotions from face or voice

Banned, unless for medical or safety reasons

2 February 2025

Images or video of real people made with AI and published

Disclosed as a deepfake

2 August 2026

An AI chat for customers on the website

People must know they talk to AI, set up by the provider

2 August 2026

Screening CVs or ranking candidates

High-risk, employment

2 December 2027

Monitoring or evaluating staff performance

High-risk, employment

2 December 2027

Scoring the creditworthiness of individuals

High-risk, essential services

2 December 2027

## Who oversees the AI, and what is recorded?

For a high-risk use, Article 26 lists the deployer's duties. For the uses in Annex III they apply from 2 December 2027.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2) In plain words, the deployer must

-   use the system as its instructions for use say;
-   give human oversight to named people who have the competence, training, authority and support for it;
-   make sure the data it feeds in is relevant and representative, where it controls that data;
-   watch how the system works, tell the provider about problems, stop using it if it presents a risk, and report serious incidents;
-   keep the logs the system generates, where they are under its control, for at least six months unless other law says otherwise;
-   tell workers' representatives and the affected employees before using such a system at work;
-   tell people that a high-risk system helps make decisions about them;
-   use the provider's information for the data protection impact assessment that the GDPR may require, and cooperate with the authorities.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)

A breach of Article 26 carries the same ceiling as Article 50, EUR 15 million or 3% of worldwide turnover.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1) The duty to keep logs covers only logs under the deployer's control. That is a practical reason to know where each AI tool runs and who holds its records.

## What will an auditor or supervisor ask to see?

The Act prescribes no single file. These records answer the articles above, and a supervisor or a customer's auditor can ask for each.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[3](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-3)

-   **A list of AI uses.** Every tool and feature in use, what it is used for, and who in the company owns it.
-   **The risk class of each use.** Banned, high-risk, transparency duty or none, with a sentence on why.
-   **A training record.** Who was trained or briefed on AI, when and on what, which the Commission names as a sensible internal record.[3](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-3)
-   **The notices.** Labels on deepfakes and on AI text published for the public, notices on emotion recognition, and from December 2027 the information to employees and to people affected by high-risk decisions.
-   **A named person for oversight.** For each high-risk use, who oversees it and with what authority.
-   **How long records are kept.** At least six months for the logs of a high-risk system under your control, or longer where other law asks.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)
-   **The supplier's documents.** The instructions for use and a clear answer on who is the provider.

## Country by country

The AI Act applies directly in every member state, but each names its own authorities and sets penalties.[1](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-1)[2](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-2)

Country

National law

Single point of contact

Czech Republic

Bill only[5](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-5)

ČTÚ, planned[6](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-6)

Denmark

Bans only[13](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-13)

Digitaliseringsstyrelsen

Finland

1377/2025[14](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-14)

Traficom

Germany

KI-MIG[11](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-11)

Bundesnetzagentur

Greece

5321/2026[16](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-16)

Data Protection Authority

Hungary

Act LXXV of 2025[8](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-8)

Science minister[9](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-9)[10](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-10)

Ireland

Act 31 of 2026[17](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-17)

AI Office

Italy

Law 132/2025[12](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-12)

ACN

Malta

L.N. 226 of 2025[18](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-18)

MDIA

Poland

Act of 3 July 2026[7](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-7)

KRiBSI

Slovenia

ZIUDHPUI[15](https://disruptivelab.dev/en/news/ai-act-for-deployers#source-15)

AKOS

## Next step

If it would help, we would be glad to sit down with you and go through the AI your company uses, sort each use into its class and note what each one needs. If you later want one AI for the whole company that your own people control and that keeps a record of every step, that is what our AI Business Harness does. Whether either is useful is entirely your decision.

[AI Business Harness](https://disruptivelab.dev/en/ai-business-harness)

## Sources

1.  1. [EUR-Lex, Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act), Official Journal L, 12 July 2024](https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng). Read on 17 November 2026
2.  2. [EUR-Lex, Regulation (EU) 2026/1744 amending Regulation (EU) 2024/1689 (Digital Omnibus on AI), Official Journal L, 24 July 2026](https://eur-lex.europa.eu/eli/reg/2026/1744/oj/eng). Read on 17 November 2026
3.  3. [European Commission, AI literacy, questions and answers (updated 20 October 2026)](https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers). Read on 17 November 2026
4.  4. [European Commission, AI Act Service Desk, timeline for the implementation of the EU AI Act](https://ai-act-service-desk.ec.europa.eu/en/ai-act/timeline/timeline-implementation-eu-ai-act). Read on 17 November 2026
5.  5. [Vláda České republiky, Výsledky jednání vlády 9. listopadu 2026](https://vlada.gov.cz/cz/media-centrum/aktualne/vysledky-jednani-vlady-9--listopadu-2026-229921/). Read on 17 November 2026
6.  6. [Ministerstvo průmyslu a obchodu, Vláda podpořila rozvoj umělé inteligence zřízením nové rady a schválením návrhu zákona (9. 11. 2026)](https://mpo.gov.cz/cz/rozcestnik/pro-media/tiskove-zpravy/vlada-podporila-rozvoj-umele-inteligence-zrizenim-nove-rady-a-schvalenim-navrhu-zakona--295239/). Read on 17 November 2026
7.  7. [Dziennik Ustaw 2026 poz. 1003, Ustawa z dnia 3 lipca 2026 r. o systemach sztucznej inteligencji](https://eli.gov.pl/api/acts/DU/2026/1003/text/T/D20261003L.pdf). Read on 17 November 2026
8.  8. [Nemzeti Jogszabálytár, 2025. évi LXXV. törvény az Európai Unió mesterséges intelligenciáról szóló rendeletének magyarországi végrehajtásáról](https://njt.jog.gov.hu/jogszabaly/2025-75-00-00). Read on 17 November 2026
9.  9. [Nemzeti Jogszabálytár, 344/2025. (X. 31.) Korm. rendelet (hatályos állapot)](https://njt.jog.gov.hu/jogszabaly/2025-344-20-22). Read on 17 November 2026
10.  10. [Nemzeti Jogszabálytár, 90/2026. (V. 13.) Korm. rendelet a Kormány tagjainak feladat- és hatásköréről (2026.11.01. állapot)](https://njt.jog.gov.hu/jogszabaly/2026-90-20-22). Read on 17 November 2026
11.  11. [Bundesgesetzblatt 2026 I Nr. 223, Gesetz zur Durchführung der Verordnung über künstliche Intelligenz vom 22. Juli 2026](https://www.recht.bund.de/bgbl/1/2026/223/regelungstext.pdf). Read on 17 November 2026
12.  12. [Normattiva, Legge 23 settembre 2025, n. 132, Disposizioni e deleghe al Governo in materia di intelligenza artificiale](https://www.normattiva.it/uri-res/N2Ls?urn:nir:stato:legge:2025-09-23;132). Read on 17 November 2026
13.  13. [Retsinformation, LOV nr 467 af 14/05/2025, Lov om supplerende bestemmelser til forordningen om kunstig intelligens](https://www.retsinformation.dk/api/pdf/249582). Read on 17 November 2026
14.  14. [Finlex, Suomen säädöskokoelma 1377/2025, Laki eräiden tekoälyjärjestelmien valvonnasta](https://www.finlex.fi/api/media/statute/898466/mainPdf/main.pdf). Read on 17 November 2026
15.  15. [Uradni list RS, št. 85/2025, Zakon o izvajanju uredbe (EU) o določitvi harmoniziranih pravil o umetni inteligenci (ZIUDHPUI)](https://www.uradni-list.si/glasilo-uradni-list-rs/vsebina/2025-01-3035). Read on 17 November 2026
16.  16. [ΦΕΚ Α΄ 114/20.07.2026, Νόμος 5321/2026, Μέτρα εφαρμογής του Κανονισμού (ΕΕ) 2024/1689](https://ia37rg02wpsa01.blob.core.windows.net/fek/01/2026/20260100114.pdf). Read on 17 November 2026
17.  17. [Houses of the Oireachtas, Regulation of Artificial Intelligence Act 2026 (No. 31 of 2026), enacted text](https://data.oireachtas.ie/ie/oireachtas/act/2026/31/eng/enacted/a3126.pdf). Read on 17 November 2026
18.  18. [Legislation Malta, L.N. 226 of 2025, Artificial Intelligence Regulations, 2025](https://legislation.mt/eli/ln/2025/226/eng/pdf). Read on 17 November 2026

## Change log

1.  17 November 2026 Published.

## Related articles

Guide · European rules

### [Mandatory e-invoicing and what to prepare](https://disruptivelab.dev/en/news/mandatory-e-invoicing)

Countries switch on structured e-invoices at different dates. What counts as one, when it applies to you and what your invoicing system must handle.

17 November 2026

Guide · European rules

### [NIS2 and what an auditor will ask to see](https://disruptivelab.dev/en/news/nis2-what-auditors-ask)

Whether NIS2 applies depends on sector and size. If it does, your systems must show who has access, what is recorded and how incidents are reported.

17 November 2026

## Company details

- **Company name:** Disruptive Lab s.r.o.
- **Registered office:** Vavrečkova 5262, 760 01 Zlín
- **Company ID (IČO):** 08240647
- **VAT ID (DIČ):** CZ08240647
- **E-mail:** office@disruptivelab.dev
- **Phone:** +420 777 702 032
